1/* 2 * Copyright (c) 2014 Advanced Micro Devices, Inc. 3 * 4 * Permission is hereby granted, free of charge, to any person obtaining a copy 5 * of this software and associated documentation files (the "Software"), to deal 6 * in the Software without restriction, including without limitation the rights 7 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 8 * copies of the Software, and to permit persons to whom the Software is 9 * furnished to do so, subject to the following conditions: 10 * 11 * The above copyright notice and this permission notice shall be included in 12 * all copies or substantial portions of the Software. 13 * 14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 15 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 16 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 17 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 18 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 19 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN 20 * THE SOFTWARE. 21 */ 22 23#include <clc/clc.h> 24#include <clc/clcmacro.h> 25#include <clc/integer/clc_abs.h> 26#include <clc/math/clc_mad.h> 27#include <clc/math/clc_subnormal_config.h> 28#include <clc/math/math.h> 29#include <clc/relational/clc_isnan.h> 30#include <clc/shared/clc_clamp.h> 31#include <math/clc_hypot.h> 32 33// Returns sqrt(x*x + y*y) with no overflow or underflow unless the result 34// warrants it 35_CLC_DEF _CLC_OVERLOAD float __clc_hypot(float x, float y) { 36 uint ux = as_uint(x); 37 uint aux = ux & EXSIGNBIT_SP32; 38 uint uy = as_uint(y); 39 uint auy = uy & EXSIGNBIT_SP32; 40 float retval; 41 int c = aux > auy; 42 ux = c ? aux : auy; 43 uy = c ? auy : aux; 44 45 int xexp = 46 __clc_clamp((int)(ux >> EXPSHIFTBITS_SP32) - EXPBIAS_SP32, -126, 126); 47 float fx_exp = as_float((xexp + EXPBIAS_SP32) << EXPSHIFTBITS_SP32); 48 float fi_exp = as_float((-xexp + EXPBIAS_SP32) << EXPSHIFTBITS_SP32); 49 float fx = as_float(ux) * fi_exp; 50 float fy = as_float(uy) * fi_exp; 51 retval = sqrt(__clc_mad(fx, fx, fy * fy)) * fx_exp; 52 53 retval = ux > PINFBITPATT_SP32 | uy == 0 ? as_float(ux) : retval; 54 retval = ux == PINFBITPATT_SP32 | uy == PINFBITPATT_SP32 55 ? as_float(PINFBITPATT_SP32) 56 : retval; 57 return retval; 58} 59_CLC_BINARY_VECTORIZE(_CLC_DEF _CLC_OVERLOAD, float, __clc_hypot, float, float) 60 61#ifdef cl_khr_fp64 62_CLC_DEF _CLC_OVERLOAD double __clc_hypot(double x, double y) { 63 ulong ux = as_ulong(x) & ~SIGNBIT_DP64; 64 int xexp = ux >> EXPSHIFTBITS_DP64; 65 x = as_double(ux); 66 67 ulong uy = as_ulong(y) & ~SIGNBIT_DP64; 68 int yexp = uy >> EXPSHIFTBITS_DP64; 69 y = as_double(uy); 70 71 int c = xexp > EXPBIAS_DP64 + 500 | yexp > EXPBIAS_DP64 + 500; 72 double preadjust = c ? 0x1.0p-600 : 1.0; 73 double postadjust = c ? 0x1.0p+600 : 1.0; 74 75 c = xexp < EXPBIAS_DP64 - 500 | yexp < EXPBIAS_DP64 - 500; 76 preadjust = c ? 0x1.0p+600 : preadjust; 77 postadjust = c ? 0x1.0p-600 : postadjust; 78 79 double ax = x * preadjust; 80 double ay = y * preadjust; 81 82 // The post adjust may overflow, but this can't be avoided in any case 83 double r = sqrt(fma(ax, ax, ay * ay)) * postadjust; 84 85 // If the difference in exponents between x and y is large 86 double s = x + y; 87 c = __clc_abs(xexp - yexp) > MANTLENGTH_DP64 + 1; 88 r = c ? s : r; 89 90 // Check for NaN 91 // c = x != x | y != y; 92 c = __clc_isnan(x) | __clc_isnan(y); 93 r = c ? as_double(QNANBITPATT_DP64) : r; 94 95 // If either is Inf, we must return Inf 96 c = x == as_double(PINFBITPATT_DP64) | y == as_double(PINFBITPATT_DP64); 97 r = c ? as_double(PINFBITPATT_DP64) : r; 98 99 return r; 100} 101 102_CLC_BINARY_VECTORIZE(_CLC_DEF _CLC_OVERLOAD, double, __clc_hypot, double, 103 double) 104#endif 105